[Rivers47]

I have a drive (Thinkpad Ultraslim DVD) that can't read the Q subchannel correctly (I think) and gives a lot of CRC errors. cdrdao gives up reading these and ends up writing pregaps equal to the track's start offset, it then subtracts the pregap from previous track's length, which causes whipper to fail with libdiscid rejecting the TOC with discid.disc.TOCError: Disc too long.

Bascially it ends up writing FILE 0 [track absolute end offset] and START track's absolute start offset.
Whipper currently doesn't handle this, but the initial fast TOC cdrdao read-toc --fast-toc actually works. The rip would work with the fix (AccurateRip completely matched), so I think the huge amount of CRC error is just noise.

The fix in toc.py

When FILE has start == 0 on trackNumber > 1:

1. Record _absoluteEnd = length - CD_LEADIN_FRAMES instead of adding length to currentLength.
2. On the paired START line, set absoluteOffset = length - CD_LEADIN_FRAMES, currentLength = _absoluteEnd - absoluteOffset, and pregapLength = 0.
3. Compute table.leadout = absoluteOffset + currentLength at the end of parse() (since totalLength is no longer a valid running sum once absoluteOffset is
   reassigned).

[github-actions]

💖 Thanks for opening your first pull request here! 💖